AI Shopping Moves Discovery, but Marketplaces Still Win Checkout

E-Commerce

AI shopping is attracting attention, but being recommended is not the same as winning the order. In October 23 coverage, PYMNTS highlights research showing that consumers retain strong marketplace preferences even when an AI assistant could buy elsewhere. The practical question for sellers is not simply whether an agent can find their products, but why a customer would choose their checkout. Coverage

What the research actually measures

PYMNTS Intelligence’s October report is based on an August 2026 survey of 2,191 U.S. consumers. In a hypothetical scenario with identical products, prices, shipping and returns, 40.2% preferred an autonomous assistant to buy from a large marketplace; 16.1% preferred a brand website. These are stated preferences, not observed autonomous-agent orders. Research and methodology

Among respondents reporting their most recent AI-assisted purchase, 59.7% said they completed it on Amazon. That figure describes the surveyed purchase subset, not Amazon’s share of all U.S. e-commerce or proof that AI caused sales to rise. The report explicitly distinguishes self-reported behavior from causal evidence. Research

The commercial incentives are particularly relevant to direct-to-consumer operators. Among marketplace shoppers, 53.9% identified a lower price or exclusive discount as a reason to switch to a brand or store, compared with 12.1% for an AI assistant handling checkout. These responses indicate potential appeal, not a guaranteed conversion lift from adding either feature. Research

Operator analysis: separate discovery from conversion

Our recommendation is to run separate discovery and purchase tests rather than treating “AI-ready” as a single investment category.

  • For discovery, review product specifications, variant names, compatibility details and availability. Make the information consistent across approved sales channels so a useful recommendation points to the right item.
  • For conversion, test the complete offer: delivered price, delivery promise, return terms and payment experience. Measure completed orders and contribution margin, not just AI referral visits.
  • For a brand website, compare a narrowly targeted offer or clearer shipping promise with an automated-checkout pilot. Include discounts, fulfillment and expected return costs before judging which option deserves more budget.
  • For marketplace listings, keep seller identity and service commitments clear. Do not assume that a preference for a marketplace guarantees a preference for your particular listing.

Keep the evidence in proportion

The survey was conducted before this week’s coverage; it is not a real-time holiday sales tally. Our takeaway is to preserve both routes to purchase: improve discoverability while earning checkout trust. An AI integration should be judged by incremental profitable orders in your own operation, not by applying a survey percentage directly to a revenue forecast.

Sources

← Back to News